Two kinds of action
Meta separates what happens on your ad from what happens off it. Impressions, link clicks, video views, reactions and shares happen on the ad itself, so they are counted directly. A purchase, booking or form submitted on your website happens off the ad, and Meta credits it to your ad only if it happened within a set number of days after someone saw or clicked the ad.
That second group is usually the one that matters to a business, and it is the one built on rules rather than direct observation.
Attribution settings change the count
Which off-ad actions get credited depends on the attribution model and settings chosen for each ad set. Meta currently offers a standard model and an incremental model, which tries to predict whether a conversion was caused by the ad. Under the standard model, Meta lists these settings for website conversions:
- Click-through: events within 1 day or 7 days after a link click.
- View-through: events within 1 day after someone saw the ad.
- Engage-through: events within 1 day after another kind of click on the ad, or a video play of a set length.
Meta itself warns that results cannot be compared across ad sets that use different attribution models, and that doing so leads to inaccurate conclusions. A report that changes attribution settings partway through, or compares campaigns set up differently, is not comparing like with like.
Why the tools disagree
It is normal for Ads Manager to show fewer conversions than Events Manager. Meta explains that Events Manager shows most events received from your site, while Ads Manager shows only the events attributed to your ads. Neither number is "wrong"; they answer different questions.
Your own records answer a third question. Orders in your store, appointments in your calendar and leads your team accepted are the closest thing to ground truth a business has. A good review sets Meta's figures beside them and explains the gap rather than picking whichever number looks best.
Too early is a real answer
A small number of conversions can swing widely from one day to the next. A test that has gathered little delivery has not proven anything yet, however exciting the first day looked. The Competition Bureau warns marketers against basing performance claims on "test results that are insignificant or based on mere chance or one-time effect". That standard is written for claims to consumers, but it is a sensible one for agency reports too.
What an honest review contains
- What changed since the last review, and why.
- What the evidence shows, with the attribution setting stated.
- How Meta's figures compare with your own records.
- What the evidence cannot show yet, and what would settle it.
- The next decision, and when it will be reviewed.
Why nobody should promise a return
Results depend on the offer, the price, the season, the market, the website, how fast enquiries are answered and many other things no agency controls. Platform reports are estimates built on attribution rules. An agency can promise a clear plan, careful testing and straight reporting. It cannot honestly promise a return on ad spend, and you should be wary of anyone who does.